We present a python library to encode peptides for machine learning applications. Non-canonical amino acids and post-translational modifications are supported.
Surprising pitfalls in common evaluation approaches for molecule libraries generated by deep learning models. Simple solutions are proposed.
Practical guidelines for training deep learning models on molecular string representations for bioactivity prediction.
DeepCocrystal is a convolutional neural network to predict co-crystal formation. SMILES augmentation is key to its development.
A novel approach to chemical language modeling. First application of structured state space sequence models (S4) to *de novo* design.
Measuring the impact of integrating molecular dynamics simulations to machine learning pipelines for bioactivity prediction.
A review of the deep learning approaches in low-data drug discovery. Future research directions are outlined.
We pharmacologically study chemical words and find that they can designate functional groups.
We present a python library to train more generalizable drug-target affinity prediction models.
A review of the deep learning approaches for structure-based drug discovery. Future research directions are outlined.